CARDIOVASCULAR MAGNETIC RESONANCE FEATURE TRACKING FOR ASSESSING MYOCARDIAL MECHANICS: FROM EXPERIMENTAL STUDIES TO CLINICAL RESEARCH

(1)

1

LITHUANIAN UNIVERSITY OF HEALTH SCIENCES MEDICAL ACADEMY

Tomas Lapinskas

CARDIOVASCULAR MAGNETIC

RESONANCE FEATURE TRACKING

FOR ASSESSING MYOCARDIAL

MECHANICS:

FROM EXPERIMENTAL STUDIES

TO CLINICAL RESEARCH

Doctoral Dissertation Biomedical Sciences,

Medicine (06B)

(2)

2

Dissertation has been prepared at the Department of Cardiology and at the Department of Cardiac, Thoracic and Vascular Surgery, Medical Academy of Lithuanian University of Health Sciences during the period of 2013–2017 and at the Department of Internal Medicine / Cardiology of German Heart Center Berlin during the period of 2016–2017.

Scientific Supervisor

Prof. Habil. Dr. Remigijus Žaliūnas (Lithuanian University of Health Sciences, Biomedical Sciences, Medicine – 06B)

Consultant

Prof. Dr. Rimantas Benetis (Lithuanian University of Health Sciences, Biomedical Sciences, Medicine – 06B)

Dissertation is defended at the Medical Research Council of the Medi-cal Academy of Lithuanian University of Health Sciences:

Chairperson:

Prof. Habil. Dr. Virgilijus Ulozas (Lithuanian University of Health Sci-ences, Biomedical SciSci-ences, Medicine – 06B)

Members:

Prof. Dr. Saulius Lukoševičius (Lithuanian University of Health Sci-ences, Biomedical SciSci-ences, Medicine – 06B)

Assoc. Prof. Dr. Juozas Kupčinskas (Lithuanian University of Health Sciences, Biomedical Sciences, Medicine – 06B)

Assoc. Prof. Dr. Sigita Glaveckaitė (Vilnius University, Biomedical Sci-ences, Medicine – 06B)

Dr. Julia Grapsa (Cleveland Clinic, Abu Dhabi, Biomedical Sciences, Medicine – 06B)

Dissertation will be defended at the open session of the Medical Re-search Council on the 13th_{of December 2017 at 10 am in the Conference}

Hall of the Department of Cardiac, Thoracic and Vascular Surgery, Medical Academy of Lithuanian University of Health Sciences.

(3)

3

LIETUVOS SVEIKATOS MOKSLŲ UNIVERSITETAS MEDICINOS AKADEMIJA

Tomas Lapinskas

MIOKARDO DEFORMAVIMOSI RODIKLIŲ

ĮVERTINIMAS ŠIRDIES MAGNETINIO

REZONANSO TYRIMU

BANDOMUOSIUOSE IR

KLINIKINIUOSE MOKSLINIUOSE

TYRIMUOSE

Daktaro disertacija Biomedicinos mokslai, medicina (06B) Kaunas, 2017

(4)

4

Disertacija rengta 2013–2017 metais Lietuvos sveikatos mokslų universiteto Medicinos akademijos Kardiologijos ir Širdies, krūtinės ir kraujagyslių chi-rurgijos klinikose ir 2016–2017 metais Berlyno širdies centre Vokietijoje.

Mokslinis vadovas

prof. habil. dr. Remigijus Žaliūnas (Lietuvos sveikatos mokslų universi-tetas, Medicinos akademija, biomedicinos mokslai, medicina – 06B)

Konsultantas

prof. dr. Rimantas Benetis (Lietuvos sveikatos mokslų universitetas, Me-dicinos akademija, biomeMe-dicinos mokslai, medicina – 06B)

Disertacija ginama Lietuvos sveikatos mokslų universiteto Medicinos akademijos Medicinos mokslo krypties taryboje:

Pirmininkas

prof. habil. dr. Virgilijus Ulozas (Lietuvos sveikatos mokslų univer-sitetas, Medicinos akademija, biomedicinos mokslai, medicina – 06B)

Nariai:

prof. dr. Saulius Lukoševičius (Lietuvos sveikatos mokslų universitetas, Medicinos akademija, biomedicinos mokslai, medicina – 06B)

doc. dr. Juozas Kupčinskas (Lietuvos sveikatos mokslų universitetas, Medicinos akademija, biomedicinos mokslai, medicina – 06B)

doc. dr. Sigita Glaveckaitė (Vilniaus universitetas, biomedicinos mokslai, medicina – 06B)

dr. Julia Grapsa (Klivlendo klinika, Abu Dabi, biomedicinos mokslai, medicina – 06B)

Disertacija ginama viešajame Medicinos mokslo krypties tarybos posėdyje 2017 m. gruodžio 13 d. 10 val. Lietuvos sveikatos mokslų univer-siteto Medicinos akademijos, Širdies, krūtinės ir kraujagyslių chirurgijos klinikos konferencijų salėje.

(5)

5

ABBREVIATIONS

2D – two-dimensional

3D – three-dimensional AF – atrial fibrillation

AHA – American Heart Association BMI – body mass index

BSA – body surface area

bSSFP – balanced steady state free precession CAD – coronary artery disease

CI – confidence interval

CMR – cardiovascular magnetic resonance CT – computed tomography

DENSE – displacement encoding imaging

EccSAX – left ventricular short-axis circumferential strain

ECG – electrocardiography

ECV – extracellular volume fraction EDV – end-diastolic volume

EF – ejection fraction

EllLAX – left ventricular long-axis longitudinal strain

ErrSAX – left ventricular short-axis radial strain

ESV – end-systolic volume FT – feature tracking

HARP – harmonic phase imaging HF – heart failure

ICC – intraclass correlation coefficient LA – left atrium / atrial

LAD – left anterior descending coronary artery LAEF – left atrial emptying fraction

LAEV – left atrial emptying volume LCx – left circumflex coronary artery LGE – late gadolinium enhancement LV – left ventricle / ventricular LVM – left ventricular mass

mDixon – magnetization prepared multiecho Dixon MI – myocardial infarction

MR – mitral regurgitation

MRI – magnetic resonance imaging

NSTEMI – non-ST-segment elevation myocardial infarction PCI – primary percutaneous intervention

(8)

8 RA – right atrium / atrial RCA – right coronary artery RV – right ventricle / ventricular SCD – sudden cardiac death SD – standard deviation SENC – strain encoding imaging SNR – signal-to-noise ratio

SPAMM – spatial modulation of magnetization SR – strain rate

SRa – left atrial active strain rate

SRccSAX – left ventricular short-axis circumferential strain rate

SRe – left atrial passive strain rate

SRllLAX – left ventricular long-axis longitudinal strain rate

SRrrSAX – left ventricular short-axis radial strain rate

SRs – left atrial total strain rate

STE – speckle tracking echocardiography

STEMI – ST-segment elevation myocardial infarction SV – stroke volume

εa – left atrial active strain

εe – left atrial passive strain

(9)

9

INTRODUCTION

The optimal diagnosis and management of cardiovascular diseases re-quire a comprehensive and precise assessment of structural and functional cardiac parameters. Therefore, noninvasive cardiac imaging has been devel-oped for providing relevant diagnostic information.

Clinical decision making and patient management in many of cardiovas-cular conditions largely rely on left ventricardiovas-cular ejection fraction (LVEF) as a primary measure of left ventricular (LV) function [1]. In addition to typical signs and symptoms, LVEF is considered to be an essential measurement in patients with heart failure (HF) [2]. Furthermore, this most important metric of cardiac function is used not only to assess treatment effect [3], but also has been proved to be a good predictor of incident HF, ventricular arrhyth-mias and sudden cardiac death (SCD) [4–6]. However, the discriminatory value of LVEF in predicting cardiovascular outcomes is limited above the LVEF of 45%, suggesting that in patients with HF with preserved LVEF, ejection fraction is a poor overall predictor of cardiovascular risk [7].

To date, cardiac wall motion analysis plays a central role for assessment of ventricular contractile function. Over the last decade, cardiovascular magnetic resonance (CMR) has evolved into the reference standard for the evaluation of cardiac morphology and function. Recently, several more ad-vanced diagnostic methods for the evaluation of myocardial contractile dys-function such as myocardial deformation imaging have become widely available and have been recognized as an important diagnostic and prognos-tic tool [8]. Parprognos-ticularly, myocardial strain and deformation parameters have been shown to be affected early in disease and therefore might represent more sensitive parameters in the setting of myocardial dysfunction when compared to conventional functional measures like LVEF [9].

CMR is a radiation-free, reference standard of assessment of cardiac anatomy and wall motion because of its excellent endocardial border defini-tion due to high spatial and contrast resoludefini-tion [10]. Although several quan-titative assessment techniques have been proposed, such as myocardial tag-ging, phase contrast velocity imatag-ging, displacement encoding (DENSE), and strain encoding (SENC) for strain analysis, these methods require addi-tional sequences and time [10, 11]. CMR feature tracking (CMR-FT) has been introduced using cine images and provides a fast and accurate assess-ment of both ventricular and atrial strains [12–14]. CMR-FT derived defor-mation parameters have been investigated in numerous cardiac diseases [8, 15, 16] and reasonable agreement of CMR-FT with other strain analysis techniques has been demonstrated for echocardiographic speckle tracking

(10)

10

[17] as well as for CMR tagging, which is considered the gold standard for myocardial strain measurement [18]. Thus, as CMR-FT is more and more evolving as an interesting tool for diagnostic and prognostic assessment in various cardiac diseases, reproducibility issues currently represent a major cause of concern.

Experimental animal models play an important role in the understanding of myocardial performance in different cardiac diseases [19]. Previous stud-ies have demonstrated the feasibility of two-dimensional (2D) speckle track-ing echocardiography (STE) to assess myocardial deformation in rodent models [20]. However, technical challenges with 2D echocardiography as limited imaging projections, small hearts and lower spatial resolution still limit its widespread use. The analysis of myocardial deformation offers quantitative assessment of early changes in LV function [21] and new find-ings in previously reported “negative” studies or studies with no obvious effect after therapy may improve our understanding of the underlying path-ophysiology. The advantages of CMR-FT comparing with myocardial tag-ging are that the technique does not require acquisition of additional images and helps to protect animals from prolonged anesthesia. In the light of the current discussion of improving animal welfare by methodological refine-ments to reduce suffering, ameliorated imaging techniques are an important tool to achieve this. As opposed to STE, only a small number of studies evaluating the inter- and intra-observer variability in small animal models using CMR tagging have been done [22, 23] and none using CMR-FT.

A recent observational CMR study reported that fat deposition can be de-tected in up to 78% of individuals who experienced myocardial infarction (MI) [27]. Lipomatous metaplasia, also termed fatty infiltration, is associat-ed with more adverse cardiac remodeling and larger infarct size [28]. More-over, it has been noted recently that these structural changes in the extracel-lular matrix of the myocardium increase the risk of ventricular arrhythmias and SCD [29]. Therefore, noninvasive detection of this myocardial remodel-ing could have great prognostic value.

Validated CMR technique using a three-dimensional (3D) single breath-hold ECG-gated and magnetization prepared multiecho Dixon (mDixon) sequence has gained increased interest over recent years. Fat-only images derived using this technique can be used to detect and quantify cardiac adi-pose tissue [30]. The proadi-posed fat-water separation (mDixon) method has a number of benefits, such as excellent discrimination between fat and water, improved diagnostic confidence and better signal-to-noise ratio (SNR). It, therefore, shows the potential to replace conventional fat imaging tech-niques, while the simultaneously acquired water-only images may be a

(11)

re-11

placement for conventional scar imaging, including late gadolinium en-hancement (LGE) [31–33].

Current evidence suggests that left atrial (LA) size and function are im-portant markers of adverse cardiac outcomes and cerebrovascular accidents [34, 35]. Thus, LA volume and function assessment may have prognostic relevance in patients with CAD. The atrium is perfused by branches of the proximal left circumflex coronary artery and depression of the chamber me-chanical performance has been observed in experimental studies during cor-onary artery occlusion. Interestingly, almost a half of patients presenting with acute MI experience acute mitral regurgitation (MR) [36] which carries additional risk of HF, atrial fibrillation (AF) and premature death [37]. Ad-vanced cardiac imaging using tissue tracking provides more information about atrial performance and allows detection of early functional changes.

(12)

12

1. THE AIM AND OBJECTIVES OF THE STUDY

1.1. The aim of the study

The aim of our study is to assess advantages and reproducibility of myo-cardial deformation parameters derived using cardiovascular magnetic reso-nance feature tracking technique in experimental animal (mouse) model and patients with acute and previous myocardial infarction.

1.2. The objectives of the study

1. To assess inter- and intra-observer agreement of CMR feature tracking derived strain measurements in mice and define study sample size neces-sary for future trials.

2. To assess changes in myocardial tissue after myocardial infarction using advanced CMR imaging (mDixon) and investigate changes in global and regional myocardial deformation in patients with old myocardial infarc-tion and lipomatous metaplasia.

3. To evaluate left atrial performance during acute ischemia and mitral re-gurgitation. Also, to assess inter- and intra-observer variability of CMR feature tracking derived left atrial strain and strain rate measurements.

1.3. Scientific novelty of the study

CMR-FT has been validated against myocardial tagging for the assess-ment of regional myocardial motion in humans. Current literature reports excellent inter- and intra-observer agreement and high inter-study reproduc-ibility of quantitative assessment of myocardial mechanics. Unfortunately, there are no studies assessing reproducibility of myocardial deformation parameters derived using CMR-FT in small animals.

Despite small number of animals included in the study we could show that myocardial deformation parameters derived using CMR-FT are highly reproducible. The most reproducible measures are global circumferential and longitudinal strain. Based on our analysis we also conclude that rela-tively small number of animals is sufficient to detect changes in strain pa-rameters. The strong advantage of the technique is that quantitative assess-ment of myocardial strain can be performed retrospectively as it requires conventional cine images only which are the part of every CMR study.

(13)

13

Recent observational studies report that fatty metaplasia is a common phenomenon in patients with previous MI which is associated with in-creased risk of life-threatening arrhythmias. Intramyocardial fat is indistin-guishable from scar or fibrosis by conventional CMR imaging. Advances in original fat-water separated (mDixon) imaging technique enables to better characterize myocardial tissue and has been proven in patients with ischem-ic and idiopathischem-ic dilated cardiomyopathy. Fat-only images derived using this technique can be used to detect and quantify cardiac adipose tissue. Un-til now, little research has focused on fat deposition in patients with old MI, especially using noninvasive cardiac imaging.

According to the results of our study, advanced CMR imaging ensures more detailed tissue characterization. Fatty metaplasia only is depicted with fat-water separated imaging without any relevant increase in scan time. Fur-thermore, we show that advanced mDixon technique is as robust as conven-tional LGE technique to detect and quantify scar tissue. On the basis of these findings we propose an updated and shortened imaging protocol for patients with previous MI that could be useful to better characterize and stratify our patients. Moreover, we conclude that fatty deposition has no impact on regional myocardial deformation in segments with LGE, but has a positive effect in myocardial segments adjacent to scar tissue.

Recently have been published several studies analyzing LA myocardial deformation changes during different cardiac diseases. There are few studies investigating LA mechanics during chronic MR, but almost none during new onset MR in patients with acute ST-segment elevation MI (STEMI). Furthermore, in our study we use novel and increasingly available imaging technique for the assessment of myocardial deformation – CMR-FT.

Based on our analysis we conclude that LA reservoir, passive and active strain are significantly higher in patients with acute MR whereas only peak early negative strain rate exhibits substantiation augmentation in STEMI patients. Moreover, we demonstrate excellent inter-observer and intra-observer reproducibility for all LA strain and strain rate values.

(14)

14

2. REVIEW OF LITERATURE

Cardiovascular diseases constitute a major global public health burden and account for about one third (30.9%) of individual mortality worldwide [38]. The optimal management of such patients is increasingly based on al-gorithms that use cutoff values or continuous variables, rather than ordinary and simple binary “yes” or “no” decision trees. Accordingly, noninvasive cardiac imaging techniques such as echocardiography and CMR imaging have been developed for providing diagnostic information that are often expressed numerically [39]. The assessment of LV function is, arguably, the most important component of a cardiac imaging study [40]. It is well known, however, that global measures such as LVEF, may not be sensitive enough to detect subtle changes in LV function, as in the case with an incip-ient disease [41, 42]. Cardiac muscle motion and deformation can be meas-ured using tissue tracking techniques such as STE and CMR-FT. Myocardi-al deformation imaging appears very promising and demonstrated ability to detect early contractile dysfunction in a majority of cardiovascular diseases.

2.1. Principles of tissue tracking technique

Tissue tracking technologies have enhanced the noninvasive assessment of myocardial deformation in experimental research and clinical practice. In echocardiography, images are characterized by the presence of speckles with a certain persistence [43, 44]. Thus, the technique is commonly re-ferred to as STE. In CMR, where tissue regions are identified by individual anatomical features, tissue tracking algorithm relies on feature tracking [17]. Importantly, the intrinsic differences in both techniques are related to what region of myocardium is tracked and may lead to differences in measure-ments.

The technology of tissue tracking falls in a general category of image post-processing methods known as optical flow [45]. The approach is com-parable to particle image velocimetry used in fluid dynamics [46, 47]. As with optical flow methods, the underlying principle is based on the recogni-tion of patterns of features or irregularities in the image to be tracked and following them in the successive images of a sequence. This feature track-ing approach can be applied to routine cine CMR acquisitions and is attract-ing the interest of many users in research and clinical practice [48].

CMR cine images are well suited for feature tracking by virtue of its rela-tively unrestricted access to large fields of view and its relarela-tively high signal to noise and contrast to noise ratios (Fig. 2.1.1). It provides the most

(15)

accu-15

rate and reproducible assessment of global atrial and ventricular volumes and function [49, 50]. Balanced steady state free precession (bSSFP) tech-nique gives good contrast between the myocardium and the blood pool at a spatial resolution of about 1–2 mm in-plane and 6–10 mm through-plane acquired during breath-holds of approximately 6–8 seconds [51]. A limita-tion is the temporal resolulimita-tion, which may not be able to resolve short-lived phases of cardiac motion. A standard CMR protocol covers the complete left and right ventricle with a stack of short-axis images with additional long-axis (two-, three- and four-chamber) views.

Fig. 2.1.1. Basic principle of CMR feature tracking. After manual contouring using a

point-and-click approach and application of tissue tracking algorithm CMR-FT software auto-matically detects endocardial borders through all cardiac phases. Cine images of LV

three-chamber view at end-diastolic (A) and end-systolic (B) cardiac phases. CMR – cardiovas-cular magnetic resonance; FT – feature tracking; LV – left ventricle; LA – left atrium; Ao –

ascending aorta

CMR does not appear to be able to distinguish features within the com-pact myocardium of the LV, presumably due to the relatively large dimen-sions of voxels and the relative homogeneity of water content and tissue properties within them. Tissue tracking in CMR is therefore most effective around endocardial borders, most of which are trabeculated. The epicardium can also be distinguished, although its clarity depends on the image proper-ties of overlying structures. It should be borne in mind that cine acquisitions are periodic. A cine loop representing an effectively averaged cardiac cycle is typically reconstructed from ECG-gated data acquired over several cycles in a breath-hold and delivering 25–50 reconstructed phases per heartbeat. The frame rate depends on heart rate and various acquisition parameters. Since MR acquisitions obtain data over several heart beats minor

(16)

beat-to-16

beat differences are smoothed out which, in combination with suboptimal temporal resolution, will obscure rapid isovolumetric phases and might lead to underestimation of displacement and strain values [52].

2.2. Definition of myocardial strain and strain rate

Strain and strain rate are measures of changes in shape and define tissue deformations. The use of these measures to describe mechanics of the heart muscle was introduced by Mirsky and Parmley in 1973 [53].

Strain is a mechanical characteristic that describes the relative change in length between two states. For an object on initial length L0 that is being

stretched or compressed to a new length L, the conventional strain is defined as:

𝜀𝜀 = 𝐿𝐿 − 𝐿𝐿_𝐿𝐿 0

0

This is known as Lagrangian strain because it refers to an initial unde-formed state [54]. The Greek letter epsilon (𝜀𝜀) is commonly used as a sym-bol for strain. The strain value is dimensionless and can be presented as a fractional number or as a percentage (by multiplying with 100). For in-stance, a fractional strain of -0.2 corresponds to a percentage strain of -20%. The strain is positive if L is larger than L0, meaning that the object has

lengthened, and negative if L is smaller and L0, meaning that the object has

shortened. If L equals L0, there has been no change in length, and the strain

is zero.

Strain can be computed taking a tissue segment of length L along any specified direction. When this segment is taken along the longitudinal direc-tion it gives the longitudinal strain, or circumferential strain when it is taken along the circumference, or it is the radial strain when length is taken over the thickness (Fig. 2.2.1) [52]. Endocardial longitudinal and circumferential strains are computed when the segment of tissue length L is taken from the endocardial border. Endocardial strain values are those more frequently used in clinical studies because they better represent the functional purpose of myocardial contraction, reducing the endocardial surface on the cavity to eject the stroke volume, and as such correlate best with volumetric meas-urements like ejection fraction [55]. Likewise, epicardial strain, which is typically lower than endocardial, can be evaluated along the epicardial bor-der, although this is rarely used.

(17)

17

Fig. 2.2.1. Planes of myocardial motion and deformation during systole. When tissue

seg-ment is taken along the longitudinal direction it gives the longitudinal strain, or circumfer-ential strain when it is taken along the circumference, or it is the radial strain when length

is taken over the thickness. Adapted according Cikes et al, Eur Heart J, 2016 [56] The strain rate is the temporal derivate of the strain and is equal to:

𝑆𝑆𝑆𝑆 = 𝑑𝑑𝜀𝜀_{𝑑𝑑𝑑𝑑}

This definition means that, whereas strain indicates the amount of defor-mation, strain rate indicates the rate of the deformation. The relation be-tween strain rate and strain can be compared to the relation bebe-tween velocity and displacement. Assuming the velocity is constant, displacement equals time multiplied by velocity. Similarly, assuming the strain rate is constant, strain equals time multiplied by strain rate. A positive strain rate means that the length of the object is increasing, whereas a negative strain rate means that the length is decreasing. If the length is constant, the strain rate is zero.

The acronym SR is commonly used to represent the strain rate. The unit of the strain rate is normally 1/sec or sec-1, which may be read as “per sec-ond”. A strain rate of -2 sec-1 applied over 1 sec would result in a relative strain of -2, and a corresponding percentage strain of -200%.

Note that, whereas the strain is a measurement of deformation relative to a reference state, the strain rate is an instantaneous measurement. There is no need to specify a reference state for strain rate, only the time of the mea-surement.

(18)

18

Because there are several definitions of strain, there are a corresponding number of similar definitions for strain rate. In particular, the derivate of

natural strain rate is:

𝑆𝑆𝑆𝑆 = 𝑑𝑑𝜀𝜀_{𝑑𝑑𝑑𝑑 =}1_𝐿𝐿𝑑𝑑𝐿𝐿_{𝑑𝑑𝑑𝑑}

where the rate of shortening of a length L of tissue is measured relative to the actual length of tissue, independently from its previous deformation his-tory. The Lagrangian strain and the natural strain rate are those more fre-quently used in cardiology.

2.3. Validation of CMR-FT as novel imaging modality

Tissue tracking can be used in the assessment of the mechanics of all cardiac chambers [57–59]. Whereas the assessment of LV and right ventri-cle (RV) deformations are of established clinical benefit, tissue tracking has been also used in studies to assess atrial deformations, mostly of the LA [60, 61].

CMR-FT was tested on simulation data from a series of artificial com-puter-generated loops that enabled testing of the image analysis procedure under simple and perfectly controlled conditions. The results of the simula-tion studies demonstrated that the errors in all cases were small for integral quantities (radius and strain) and slightly larger for differential quantities (velocities and strain rate) as a result of the latter being a derivate of the former. These studies demonstrated that the quality of the results had lower spatial resolution and that higher frame rates (above 25 Hz) were not bene-ficial if not accompanied by an increase in spatial resolution [62].

CMR-based myocardial line tagging was probably the first technique to assess regional myocardial deformation noninvasively [63]. The original linear tagging approach was then augmented by two orthogonally intersect-ing sets of lines mark rectangular grids in a 2D image by Spatial Modulation of Magnetization (SPAMM or C-SPAMM) techniques (Fig. 2.3.1). These have been considered the gold standard for myocardial strain, although their limitations include the possible obscuring by tag lines of some endocardial borders, suboptimal temporal resolution and the additional, dedicated acqui-sition time. The methods of quantifying deformation from tagged images are essentially similar to those of feature tracking [64].

In CMR-FT the most consistent parameter is global circumferential strain, closely followed by global longitudinal strain [65, 66]. In contrast, variations in global radial strain between studies are large. Circumferential

(19)

19

strain may present improved reproducibility in CMR-FT than in STE pre-sumably for the higher quality of CMR short-axis images. The high echo-genicity of the fibrous annulus enables more accurate results for longitudi-nal strain by STE [67–69].

Fig. 2.3.1. Myocardial tagging in LV short-axis at end-diastolic (A) and end-systolic (B)

cardiac phases using spatial modulation of magnetization (SPAMM) showing no defor-mation of tagging grid during ventricular systole in mid-ventricular anterior, anteroseptal

and inferoseptal segments [70]

Global circumferential strain from the mid LV slice has been shown to correlate strongly with myocardial tagging using Harmonic Phase Imaging (HARP) in patients with Duchene Muscular Dystrophy and varying LV dys-function [17]. Global circumferential strain has also been compared with a C-SPAMM myocardial tagging reference method and revealed that agree-ment was modest and notably segagree-mental strain measureagree-ments were less ac-curate than global measurements [11]. However, comparability is still in-conclusive in pathologic conditions and non-satisfactory for regional defor-mation.

DENSE is another technique for strain calculation, first introduced in 1999 [71]. The phase-reconstructed images by DENSE have high temporal and spatial resolution, and direct extraction of motion data [72]. However, this technique requires some specific sequences, which may prolong scan time. So far, applications of DENSE in clinical practice have been still lim-ited by small sample size and only applied for short-axis images. The pooled normal ranges of DENSE are lower than for CMR-FT.

SENC was developed on the concepts of myocardial tagging, but it uses tag planes parallel to the image plane [73]. In other words, LV global longi-tudinal strain is obtained from short-axis views, and LV global

(20)

circumferen-20

tial strain from long-axis views. Interestingly, SENC can provide RV global circumferential strain.

2.4. Normal values for CMR-FT

Across normal value studies, the most consistent parameters are global circumferential and global longitudinal strain while variations in global ra-dial strain between studies are large and rara-dial strains measured from long-axis acquisitions are often lower than from short-long-axis acquisitions, possibly attributed to through-plane motion in the short-axis scans. Normal values do not depend on field strength when acquired with similar acquisition parame-ters including similar spatial resolution [74]. Comparative normal values can be found in Table 2.4.1.

Table 2.4.1. Overview of normal values for CMR-FT

First author, Year (Ref. #) n EccSAX ErrSAX ErrLAX EllLAX

Augustine et al, 2013 (18) 145 -21 ± 3 25 ± 6 – -19 ± 3 Andre et al, 2015 (75) 150 -27 ± 4 36 ± 8 – -23 ± 3 Kutty et al, 2013 (76) 20 -25 ± 2 50 ± 12 – -20 ± 5 Kempny et al, 2012 (15) 26 -24 ± 6 28 ± 11 – -21 ± 3 Morton et al, 2012 (65) 10 -17 ± 5 19 ± 7 18 ± 6 -20 ± 5 Padiyath et al, 2013 (69) 20 -25 ± 3 51 ± 12 – -20 ± 5 Schuster et al, 2011 (77) 10 -24 ± 7 20 ± 15 15 ± 10 -16 ± 10 Taylor et al, 2015 (40) 100 -26 ± 4 40 ± 8 – -21 ± 5 Values are expressed as n or mean ± standard deviation. EccSAX – left ventricular short-axis

circumferential strain; ErrSAX – left ventricular short-axis radial strain; ErrLAX – left

ventric-ular long-axis radial strain; EllLAX – left ventricular long-axis longitudinal strain.

RV circumferential and radial strain and strain rate values showed overall lower values than their LV counterparts did [16]. This is consistent with RV geometrical features, having thinner walls and larger radius of curvature of the free wall.

Normal values around -20% for the LV and lower than -20% for the RV have been recommended for clinical practice. However, radial strain shows large ranges between studies and variability of segmental strains remains too high to use them as single clinical measures.

Normal values for LA global longitudinal strain can be derived during the reservoir (29 ± 5%), conduit (21 ± 6%), and atrial contraction phases (8

(21)

21

± 3%) of the atrial deformation and show an increase in atrial contraction in elderly subjects consistent with physiology of normal aging [78, 79].

2.5. CMR-FT reliability and reproducibility

Objective and reproducible methods to quantify myocardial function are of great importance for patient management, therapy monitoring and out-come studies. Myocardial strain assessed by CMR-FT is an increasingly used, fast and cost efficient way of determining early changes of alterations in cardiac function during disease [75].

In a recent study Schmidt et al. elegantly demonstrated that LV global and regional circumferential as well as longitudinal strain and SR parame-ters are highly reproducible on both, the intra- and inter-observer level [80]. A similar excellent intra- and inter-observer agreement of CMR-FT derived global circumferential and longitudinal strain has been shown previously in several studies [15, 18, 81–84], whereas the reproducibility of global cir-cumferential and longitudinal SR has been tested in only a minority of available CMR-FT studies [18, 83]. Based on all those studies, global cir-cumferential and longitudinal strain and SR parameters can be reliably used in future studies.

Only few studies so far investigated the reproducibility of regional LV strain parameters [18, 85]. Regional strain parameters are of particular im-portance, as global values have been shown to not fully reflect the degree of regional wall motion abnormalities and, moreover, the limited reproducibil-ity of segmental strain measurements is a known issue [65, 75]. Hence, strain analysis on a regional rather than a global or segmental level might represent a compromise in situations where regional strain abnormalities are expected. Despite an overall excellent reproducibility of regional strain and SR parameters, it is noteworthy that apical parameters are less reproducible when compared with other regional parameters [80]. One potential explana-tion for this finding might be a higher degree of myocardial trabeculaexplana-tions in the apical regions of the LV, altering voxel appearances and thereby mak-ing accurate trackmak-ing difficult [18].

Global as well as regional radial strain and SR parameters demonstrate a considerably lower intra- and inter-observer reproducibility [80]. Some au-thors speculate that the reasons for this effect might be due to geometry of the heart with analysis in a plane of movement with the smallest potential diameter for tracking [18], while others propose that the cause of a lower reproducibility of radial strain might be problems concerning the measure-ment of an interaction of endocardial and myocardial borders during the

(22)

22

tracking, which is not necessary for the derivation of longitudinal and cir-cumferential strain parameters [57]. In addition, changes in the voxel pattern within the myocardium during the cardiac cycle as well may have an impact on the consistency of the feature tracking algorithm and account for some variation particularly for radial strain measurements [80]. Irrespective of the underlying cause, radial strain and SR parameters derived from single surements are not recommended in clinical studies. If radial strain mea-surements are considered to be important parameters, it might be favorable to use an average of three analysis repetitions [66]. In addition, studies in-cluding radial strain measures should be based on appropriate sample size considerations [80].

2.6. Sample size calculation

High reproducibility (low standard deviation) leads to greater reliability of observed changes in a parameter under measure, and smaller sample sizes in clinical trials. The ratio of a standard deviation of inter-study reproduci-bility between techniques is related to sample size by a square function, and therefore relatively small improvements in reproducibility have greatly magnified effects on sample size. As the study conducted by Grothues et al. has shown CMR to have considerable improvements in inter-study repro-ducibility compared with echocardiography, the sample size reductions through its use compared with echocardiography are substantial, ranging from 55% to 93% [86].

The impact of sample size reduction lies in faster study completion and the potential for reduced costs, particularly for the pharmaceutical compa-nies conducting clinical trials. The relatively small increment in costs from the use of CMR is outweighed by substantial cost savings from the smaller number of patients in a study group from 3 areas: reduced hospital and phy-sician performance fees; reduced internal company costs of organizing and monitoring studies; and highly significant savings if drugs can be brought to the market faster [86].

2.7. Value of experimental studies

Small animal models play a vital role in the study of cardiac disease mechanisms, and there is a great need for methodology for evaluation of in

vivo myocardial function in animals. Several methods exist for assessment

of myocardial deformation in small animals, including ultrasound speckle tracking and tissue Doppler, as well as a variety of CMR-based techniques.

(23)

23

Challenges with 2D echocardiography include restrictions in available im-aging projections and limited spatial resolution. CMR provides superior ge-ometric freedom and thus allows standardized projections, regardless of heart geometry [87]. This is an absolute requirement when assessing region-al function over time, for instance when evregion-aluating disease progression. Phase-contrast MRI (PC-MRI) offers measurement of the geometry, motion, and strain of the myocardium with high spatial and temporal resolution [88– 90].

Rodents are widely used as an animal model for investigating cardiovas-cular disease, such as HF, acute MI, and hypertension [91–93]. Many stud-ies have used M-mode and 2D echocardiography for the assessment of ven-tricular function in rats [94, 95]. Although these techniques are adequate in the absence of regional wall motion abnormalities, the small size of the rat heart and the relatively fast heart rate preclude accurate measurements in case of regional wall motion abnormalities. Doppler-derived strain meas-urements is a valuable technique for the quantification of regional myocar-dial contractility that has been validated in a variety of experimental animal models and clinical settings [96–98]. However, technical issues, including angle dependence, signal noise, and measurement variability, may limit its routine use in rodent models.

It is now understood that the regional cardiac performance changes earli-er than the global function. Reports from sevearli-eral settings showed that myo-cardial strain and SR have the potential to discriminate viable from infarcted myocardium [99, 100]. After infarction, the velocity of cardiac contractility decreases, which is reflected by the reduction of strain. Derumeaux et al. have found that Doppler myocardial imaging could differentiate non-transmural from non-transmural infarction in open-chest pigs study [101]. Other study reported that the peak systolic radial and circumferential strain and SR enable the distinction between normal, hypokinetic and akinetic segments as defined by CMR [102].

Adverse LV remodeling occurs within hours following MI and is one of the most important prognosticators for short and long-term survival. HF is a result of progressive cardiac enlargement and dysfunction, processes that continue months to years after the event. Two fundamental processes that occur after a larger anterior MI and that are responsible for the occurrence of HF, arrhythmias and death are replacement fibrosis and LV dilatation through adverse remodeling. In theory, replacement fibrosis affects strain through loss of contractile elements, whereas remodeling affects cardiac function through increased stress and decreased contractility of remaining myocardium [103]. The assessment of alterations of regional myocardial function and LV remodeling after MI is very important for prognosis and

(24)

24

therapy, including revascularization and stem cell therapy. In evolving in-farcts, myocardial fibrosis is increased along with LV remodeling, and re-duced tethering between myofiber bundles, leading to impaired cardiac function [104]. So, myocardial strain assessment can be utilized as a nonin-vasive tool for monitoring changes in regional function after therapeutic interventions [105].

2.8. Prevalence and prognostic significance of myocardial fat

Fatty infiltration within the myocardium may appear in isolation or in as-sociation with fibrous tissue [106, 107]. Intramyocardial fat is correlated with various cardiomyopathies, but is also present to some degree of normal myocardium. The pathophysiology of fibrofatty infiltration is not fully un-derstood. Fatty infiltration of fibrous tissue is observed in imaging studies done by both CT [108, 109] and MRI [110–112], and has been reported in a number of histology studies of samples from endocardial biopsies as well as autopsies [103, 104]. The mechanism of fatty infiltration, referred to as li-pomatous metaplasia, has been described by several authors [26]. The measurement of fibrofatty infiltration by imaging studies is relatively new.

Histological evidence of fibrofatty infiltration is a hallmark of ar-rhythmogenic right ventricular dysplasia, and is also evident in previous MI and other nonischemic cardiomyopathies. A higher frequency of fatty tissue is found in obese people and heavy beer drinkers. It has been shown that fibrofatty infiltration of the myocardium is associated with SCD, and there-fore noninvasive detection could have high prognostic value [29].

Conventional approaches to fat and water discrimination based on fat suppression are commonly used to characterize masses, but have reduced ability to characterize myocardial fatty infiltration due to the poor contrast of microscopic fat and partial-volume effects. Multi-echo Dixon-like meth-ods for fat and water separation provide a sensitive means of detecting small concentrations of fat with improved contrast [115–119]. The multi-echo approach to water and fat separation has a number of advantages over chem-ical-shift suppression: 1) fat has positive contrast; 2) water and fat images can be acquired in a single acquisition avoiding spatial misregistration; 3) the method is compatible with pre-contrast and late-enhancement imaging; 4) it is less susceptible to partial-volume effects and 5) robust in the pres-ence of background field variation; and 6) it eliminates chemical-shift arti-fact [120].

After LV MI, changes in tissue composition and the structure of the heart determine cardiac function, as well as future cardiac events and mortality

(25)

25

[121]. Well-known changes include the formation of scar tissue and cardiac remodeling [122]. Lesser known is the association of increased fat content with areas of healed LV infarction. Whether it is termed accumulation [123], metaplasia [26], deposition or infiltration [27], intramyocardial fat may cause rupture [124] or affect the conduction system [125]. Past autopsy studies have shown that 68% and 84% of LV myocardial scars associated with chronic ischemic heart disease showed fatty replacement [126] and a recent observation imaging study showed that 78% of LV infarctions demonstrated fat deposition [27].

The presence of lipomatous metaplasia appears to be a stronger predictor than infarct size alone and remains significant when added to LV volumes or ejection fraction. A recent study in an animal model conducted by Pouli-opoulos et al. found that the presence of fatty infiltration in sheep with ex-perimentally induced MI was associated with more inducible ventricular tachycardia. The presence of myocardial fat has also been associated with inducible tachycardia in humans in the RV. It may be that adipose tissue predisposes to re-entrant tachycardias and impairs myocardial conduction, leading to increased risk of ventricular arrhythmias and death [127].

2.9. Left atrium and coronary artery disease

The LA exerts multiple functions, acting as a reservoir during LV systo-le, as a conduit for blood flowing from the lungs to the LV cavity during diastole, as an active contractile chamber after atrial electrical activation and as a volume sensor for the heart by releasing natriuretic peptides [128, 129].

In recent years, a growing body of evidence has developed suggesting that LA size and function are markers of adverse cardiovascular outcomes and cerebrovascular accidents [35]. During ventricular diastole, the atrial cavity is directly exposed to LV pressures. With worsening LV compliance, atrial pressure increases to maintain adequate LV filling, which results in LA enlargement [130]. Therefore, LA volume may potentially reflect the severity of the underlying diastolic LV dysfunction. Furthermore, the com-prehensive analysis of regional LA function may add information about atrial electromechanical remodeling, being useful for prognostic stratifica-tion, AF risk and management [131].

Acute MI can modulate not only systolic, but also diastolic LV function. The important consequence of diastolic dysfunction is the elevation of LV filling pressures [132]. LV diastolic dysfunction has been related to morbid-ity and death independently of systolic function derangement in acute in-farction [133, 134].

(26)

26

A number of studies have shown that LA volume and function predicts survival after an acute MI. Jing et al. measured LA and LV global longitu-dinal strain rate by 2D STE in 50 patients with non-ST-segment elevation MI (NSTEMI) and 40 age-matched controls [135]. Compared with healthy subjects, NSTEMI patients had significantly decreased LA global longitudi-nal SR. Global LA peak negative strain rate during early ventricular diastole was significantly correlated with LV contractile function. Kühl et al. meas-ured LA size and function from coronary CT in 384 NSTEMI patients. Af-ter adjustments for age, number of diseased coronary arAf-teries, LVEF and Killip class, both LA fractional change and atrial pump function resulted as significant independent predictors of all-cause mortality, but not LA maxi-mal volume [136].

In 843 patients, peak LA longitudinal strain, measured within 48 h of MI admission, was significantly related to the composite outcome of death and congestive HF during a follow-up lasting for a median of 23 months [137]. However, this effect did not persist after adjustments for global LV dinal strain, maximal LA volume and age, suggesting that peak LA longitu-dinal strain represents a composite of longitulongitu-dinal LV systolic function and maximal LA volume.

2.10. Left atrial function in patients with mitral regurgitation

In chronic degenerative MR, the LV and LA are subject to increased pre-load. In previous animal studies, it has been shown that in the compensated phase of chronic MR, the LA becomes larger and more compliant with a more potent booster pump action as a result of the optimal use of the Frank-Starling mechanism of LA muscle [138]. In accordance with animal studies, Frank-Starling behavior of the LA has been demonstrated in clinical studies including different patients with MR. As MR progressively increased, the extent of LA shortening and expansion is diminished, indicating that the extreme dilation no longer provokes the Frank-Starling response because the LA myocardium may operate on the descending limb of the Frank-Starling curve [139].

Increased LA mechanical work in chronic MR may contribute to LA failure and AF [140, 141]. Atrial contractile performance improves after surgery, suggesting a state of decreased atrial function prior to surgery, even in patients with apparently normal LV function [142]. LA enlargement is accompanied by chronic inflammatory changes, cellular hypertrophy, and intersticial fibrosis that increase vulnerability to AF [143, 144]. The pres-ence of AF and LA dilatation in patients with MR are associated with

(27)

in-27

creased cardiovascular morbidity and mortality and an increased risk of SCD [37, 145, 146].

LA function indices such as strain and SR have been proposed using non-invasive imaging modalities such as STE and color tissue Doppler imaging [147–149]. CMR has been reported as the standard of reference for LV mass and volumes as well as systolic function evaluation. However, only few studies have been dedicated to LA geometry and volumes, highlighting the higher accuracy of CMR measurements [150, 151]. A growing body of literature suggests to focus on the quantification of the three basic LA func-tions rather than on the LA volumes only: LA reservoir function has shown to closely correlate with LV filling pressures [152] and has demonstrated to be a sensitive biomarker for the prediction of adverse cardiac events inde-pendently of other measures of cardiac dysfunction in patients with HF [153]. LA physiology and pathophysiology as quantified by STE and CMR-FT carry promising clinical and prognostic implications. Future studies will need to apply LA deformation imaging to support our understanding of HF development and risk stratification in valvular heart disease, AF, hyperten-sion, CAD and cardiomyopathies [154].

(28)

28

3. MATERIALS AND METHODS

3.1. Reproducibility and sample size calculation in mice

This preliminary study was performed to evaluate inter- and intra-observer reproducibility of CMR-FT derived strain measurements in a small animal (mouse) model. Also, we calculated the necessary study sample size to define the number of animals required for future studies.

3.1.1. Animals

Six C57BL/6J male, mice were randomly selected from the ongoing ex-perimental mouse model at German Heart Center Berlin. Animals used in this study were maintained in accordance with Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health [155]. All animal procedures comply with the guidelines of the German Law on the Protection of Animals and the protocol was approved by an in-stitutional review committee (G0099/14).

3.1.2. Cardiac magnetic resonance

All CMR measurements were performed on a 3 Tesla small animal MRI system (MR Solutions, Guildford, United Kingdom) with a quadrature bird-cage cardiac volume coil. After induction of inhalative anesthesia using isoflurane-oxygen (4–5%), animals were placed on a dedicated mouse sledge and MR-compatible ECG electrodes were attached to the paws. An-esthesia was maintained with isoflurane-oxygen (1.5–2%) to adjust heart rate at 400–450 beats per minute. Images were acquired using respiratory and ECG-gated gradient-echo cine sequences in two-chamber long-axis, four-chamber long-axis and five to seven short-axis planes completely cov-ering the LV (Fig. 3.1.2.1). Relevant acquisition parameters included: 15 phases per cardiac cycle, repetition time (TR) 10 ms, echo time (TE) 3 ms, averages 4, field of view (FOV) 40 × 40 mm, pixel size 0.15 × 0.15 mm, slice thickness 1 mm. All animals underwent two CMR examinations with a four-week time interval between each study.

3.1.3. LV volumetric and functional analysis

Volumetric analysis was performed offline using commercially available software CMR42 (Circle Cardiovascular Imaging Inc., Calgary, Canada). LV end-diastolic (LVEDV) and end-systolic (LVESV) volumes were quantified using manual planimetry of the endocardial and epicardial surface from

(29)

29

short-axis stack and LV stroke volume (LVSV), LVEF, myocardial mass and cardiac output were calculated.

Fig. 3.1.2.1. Example of cine images in mouse. Four-chamber at diastolic (A) and

end-systolic (B), two-chamber long-axis at end-diastolic (C) and end-end-systolic (D), and mid-ventricular short-axis at end-diastolic (E) and end-systolic (F) cardiac phases. LV – left

ventricle; LA – left atrium; RV – right ventricle; RA – right atrium

3.1.4. Myocardial feature tracking

The cine images were used to calculate myocardial strain and strain rate offline using dedicated software (TomTec Imaging Systems, 2D CPA, MR, Cardiac Performance Analysis, Unterschleissheim, Germany). Endocardial and epicardial contours were manually drawn in both long-axis and one

(30)

30

mid-ventricular short-axis views at end-diastole for each mouse by two in-dependent observers. After application of a tracking algorithm the software automatically identified endocardial borders throughout the cardiac cycle and computed mean segmental and global myocardial strain and strain rate parameters. All images were analyzed three times and derived measure-ments were averaged. LV global longitudinal strain and strain rate (EllLAX

and SRllLAX) were calculated by averaging the strain curves of both

two-chamber and four-two-chamber long-axis views (three measurements in two im-aging planes of two scans in six mice resulted in 72 measurements) whereas global circumferential and radial strain (EccSAX and ErrSAX) and strain rate

(SRccSAX and SRrrSAX) were derived using one mid-ventricular short-axis

view containing both papillary muscles (three measurements in one imaging plane of two scans in six mice resulted in 36 measurements) (Fig. 3.1.4.1). To assess intra- observer agreement data analysis was repeated 4 weeks af-ter initial assessment.

Fig. 3.1.4.1. Cine CMR images with endocardial contouring and examples of CMR-FT

myocardial strain curves in mouse. A frame from cine image of the LV two-chamber view after application of tissue tracking algorithm (A) depicting LV and LA and example of LV global longitudinal strain curve (B). Cine image of LV short-axis view at mid-ventricular level after automatic endocardial border detection (C) and example of LV global radial

strain (D). LV – left ventricle; LA – left atrium; RV – right ventricle RA – right atrium; CMR-FT – cardiovascular magnetic resonance feature tracking

(31)

31

3.2. Fatty metaplasia quantification and impact on regional myocardial deformation

The main goal of this study was to assess the advantages of recent CMR imaging techniques. We also investigated changes in myocardial tissue after MI using mDixon technique. In addition, we evaluated changes in global and regional myocardial deformation in patients with lipomatous metaplasia using CMR-FT.

3.2.1. Study population

We retrospectively enrolled 20 subjects with old MI (infarct age median, 60.0 months; range, 13.0–90.0 months) to assess global and regional cardiac function, myocardial tissue composition, viability and new myocardial is-chemia. The medical history of the participants was obtained from medical records. The study complies with the Declaration of Helsinki and was per-formed in accordance with local law. Inper-formed consent was obtained from all patients.

All CMR images were acquired using 1.5 (Achieva) or 3 Tesla (Ingenia, Philips Healthcare, Best, The Netherlands) MRI scanners with a 32-channel cardiac surface coil in supine position. All study participants were scanned using an identical comprehensive imaging protocol.

The study protocol included initial scouts to determine cardiac imaging planes. Cine images were acquired using ECG-gated bSSFP sequence with multiple breath-holds at end-expiration in three LV long-axis (two-chamber, three-chamber and four-chamber) planes. The ventricular two-chamber and four-chamber planes were used to plan the stack of short-axis slices cover-ing the entire LV. The followcover-ing imagcover-ing parameters were used: for 1.5 T scanner: repetition time (TR) = 3.3 ms, echo time (TE) = 1.6 ms, flip angle = 60°, acquisition voxel size = 1.8 × 1.7 × 8.0 mm3_{and 30 phases per}

cardi-ac cycle; for 3 T scanner: TR = 2.9 ms, TE = 1.45 ms, flip angle = 45°, cardiac-quisition voxel size = 1.9 × 1.9 × 8.0 mm3 and 30 phases per cardiac cycle.

The LGE images were obtained 10 minutes after the injection of 0.15 mmol/kg gadobutrol (Gadovist®, Bayer Schering Pharma AG, Berlin, Ger-many). A Look-Locker sequence was acquired to determine the inversion time to null the signal of the LV myocardium. A 3D inversion recovery fat saturated spoiled gradient echo sequence was used to detect scar tissue in three LV long-axis and short-axis orientations (Fig. 3.2.2.1). Typical param-eters for imaging were: for 1.5 T scanner: TR = 3.3 ms, TE = 1.6 ms, flip

(32)

32

angle 15°, acquisition voxel size 1.6 × 1.6 × 10.0 mm3_{, parallel imaging}

fac-tor (SENSE) = 2.2; for 3 T scanner: TR = 3.2 ms, TE = 1.62 ms, flip angle 15°, acquisition voxel size 1.5 × 1.5 × 10.0 mm3, SENSE = 2.2. The breath-hold time varied from 12 to 16 seconds, depending on anthropometrics. Single 3D long-axis scan was acquired during one hold. Two breath-holds were required to acquire full 3D short-axis scan.

Fig. 3.2.2.1. Example of patient with history of two previous MI. Conventional LGE (A),

in-phase (B), fat-only (C) and water-only (D) mDixon images acquired in ventricular four-chamber view. LGE, in-phase and water-only mDixon images demonstrate scar tissue in the medial inferoseptal and medial anterolateral segments (arrows). Fat deposition is

visi-ble in fat-only image in corresponding myocardial segments (arrows). MI – myocardial infarction; LGE – late gadolinium enhancement; mDixon – multiecho inversion recovery

spoiled gradient echo

Additionally, in all study participants a single breath-hold ECG-gated 3D inversion recovery spoiled gradient multiecho (mDixon) sequence was used for fat water separation imaging. The following sequence parameters were used: for 1.5 T scanner: TR = 4.7 ms, TE1 = 1.5 ms, TE2 = 3.0 ms, flip an-gle = 15°, voxel size 1.6 × 1.6 × 5.0 mm3, parallel imaging factor (SENSE) = 2.2; for 3 T scanner: TR = 3.8 ms, TE1 = 1.3 ms, TE2 = 2.4 ms, flip angle = 12°, voxel size 1.7 × 1.7 × 5.0 mm3, SENSE = 2.2. In-phase, fat-only and water-only images (Fig. 3.2.2.1) were reconstructed at the scanner.

3.2.3. Image analysis

All images were analyzed offline using commercially available software (Medis Suite, version 2.0, Leiden, The Netherlands) in accordance to a re-cent consensus document for quantification of LV function and mass using CMR [156]. LVEDV and LVESV volumes were quantified using manual planimetry of the endocardial and epicardial surface from short-axis stack and LVEF, myocardial mass, cardiac output and cardiac index were calcu-lated. Papillary muscles were considered part of the blood pool. LV

(33)

vol-33

umes and myocardial mass were adjusted to body surface area which was calculated by the Mosteller method.

The endocardial and epicardial contours drawn on cine images were transferred into LGE images. The presence and extent of LGE were quanti-fied using the Signal Threshold versus Reference Mean (STRM) >3 stand-ard deviations (SD) method as it provides the greatest accuracy with ac-ceptable reproducibility compared with other signal intensity threshold techniques [157]. All algorithm-selected pixels in the myocardium were counted on each of the LGE images. The total LGE volume and mass were calculated automatically. The extent of scar tissue was defined using a 5-point scale where 0 = absence of LGE; 1 = LGE of 1 to 25% of LV wall thickness; 2 = LGE extending to 26–50%; 3 = LGE extending to 51–75%; and 4 = LGE extending to 76–100% [158].

The endocardial and epicardial contours were similarly transferred into mDixon images using the same program. The fat deposition volume and mass were calculated using the same signal intensity threshold level (>3 SD) in the fat-only images. The extent of fat was classified as subendocar-dial, subepicardial and transmural. Fibrosis volume and mass were calculat-ed using the identical approach in the water-only images. Additionally, the extent of LGE was assessed in the in-phase mDixon images. The location of LGE and fat deposition was defined using a standard American Heart Asso-ciation (AHA) 17-segment model [159]. The global LGE (fibrosis plus fat) and fibrosis as percentages of LV mass and the fatty metaplasia as percent-age of LV mass and global LGE mass were calculated.

The cine images were used to calculate circumferential (EccLV) and

radi-al (ErrLV) strain using commercially available software CMR42 (Circle

Car-diovascular Imaging Inc., Calgary, Canada). LV endocardial and epicardial borders were contoured by a point-and-click approach in three short-axis slices (basal, mid-ventricular and apical) at LV end-diastolic phase. After application of a tissue tracking algorithm endocardial and epicardial borders were detected through all cardiac phases. The right ventricular upper septal insertion point was manually defined to allow accurate segmentation ac-cording to an AHA 16-segment model. LV strain analysis was performed on a segmental level. Segments with LGE extent of <50% of LV wall thickness were excluded from analysis, because the sensitivity and specificity of the method to detect myocardial segments with scar is highest when LGE extent is >50% of LV wall thickness [160]. All image analysis was performed by two experienced (CMR level 3) investigators.

(34)

34

3.3. LA mechanics in patients with STEMI and MR

We performed this study to investigate LA myocardial performance dur-ing acute ischemia and subsequent volume overload due to MR. Additional-ly, we evaluated inter-observer and intra-observer reproducibility of CMR-FT derived LA strain and SR measurements.

3.3.1. Study population

Patients were consecutively enrolled into the study if they presented with first STEMI and were treated with primary coronary intervention (PCI) at Hospital of Lithuanian University of Health Sciences (HLUHS). The diag-nosis of STEMI was based on typical symptoms, specific electrocardio-graphic (ECG) changes (ST-segment elevation greater than 1 mm in two contiguous limb leads or more than 2 mm in precordial leads or new left bundle branch block), elevated troponin levels and detection of occluded coronary artery during conventional coronary angiography.

Transthoracic two-dimensional echocardiography was performed within 72 hours from admission and primary PCI. The severity of MR was as-sessed according to the European Association of Cardiovascular Imaging (EACVI) recommendations by proximal isovelocity surface area (PISA) method or semi-quantitative color flow Doppler when quantitative assess-ment of MR was not feasible (unmeasurable PISA or continuous Doppler trace) [161]. Previously has been proved that only mild secondary MR has impact on LA longitudinal deformation therefore we selected STEMI pa-tients with mild-to-moderate MR [162]. Ischemic MR is highly load de-pendent. There was no significant difference on hemodynamic conditions measured by echocardiography. According to our findings patients were di-vided into two following groups: patients without MR (MR(-)) or with func-tional MR (MR(+)). Patients with trace MR were considered as MR(-). The final study population consisted of 20 STEMI patients: 10 without and 10 with secondary MR.

Exclusion criteria were: medical history of ischemic heart disease (known coronary artery disease, previous MI, PCI, coronary artery bypass grafting), structural cardiac valve disease (including previous valvular sur-gery), previously known MR. We excluded subjects with anterior STEMI to avoid distinct ischemic effect on LA myocardial performance. Patients with absolute contraindications for CMR (ferromagnetic implants, vascular aneu-rysm clips or claustrophobia) and those with irregular heart rhythm (multi-ple premature beats or atrial fibrillation) were also excluded. The study complies with the Declaration of Helsinki and was approved by local Ethics

(35)

35

Committee. All patients gave written informed consent before entering the study.

All CMR images were acquired using 1.5 Tesla scanner (Siemens Mag-netom Aera, Siemens AG Healthcare Sector, Erlangen, Germany) with an 18-channel phased array coil in a supine position. CMR was performed within 72 hours from echocardiography. The study protocol included an ini-tial survey to define imaging planes. Cine images were acquired using retro-spectively gated bSSFP sequence with short periods of breath-holding in three LV long-axis planes. The ventricular two-chamber and four-chamber planes were used to plan contiguous stack of short-axis slices covering en-tire LV. The in-plane resolution of cine images was 0.9 × 0.9 mm, slice thickness of 8 mm with 2 mm interslice gap and 25 phases per cardiac cy-cle.

3.3.3. Volumetric analysis

Volumetric analysis was performed using dedicated software (Syngo.via, Siemens AG Healthcare Sector, Erlangen, Germany). LVEDV and LVESV were quantified using manual planimetry of the endocardial and epicardial surface from short-axis stack as previously described and LVEF was calcu-lated. Papillary muscles were considered as being part of the blood pool. Ventricular volumes were adjusted to body surface area. To assess phasic atrial function by the volumetric method, LA volumes were estimated using the previously validated biplane-area method according to the formula [163]:

𝐿𝐿𝐿𝐿 𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣 (𝑣𝑣𝑣𝑣) = 0.85 × 𝐿𝐿𝐿𝐿 𝑎𝑎𝑎𝑎𝑣𝑣𝑎𝑎2𝑐𝑐ℎ × 𝐿𝐿𝐿𝐿 𝑎𝑎𝑎𝑎𝑣𝑣𝑎𝑎4𝑐𝑐ℎ / 𝐿𝐿𝐿𝐿 𝑣𝑣𝑣𝑣𝑙𝑙𝑙𝑙𝑑𝑑ℎ

LA length was defined as the longest distance measured between posteri-or atrial wall and mid-pposteri-ortion of mitral annulus plane in either the two-chamber or four-two-chamber view. LA volumes were estimated at three differ-ent time points of the cardiac cycle: LA maximal volume was measured dur-ing ventricular end-systole at the phase of mitral valve opendur-ing, LA minimal volume was obtained at late ventricular end-diastole after LA contraction and mitral valve closure and LA volume at the onset of atrial systole was considered at left ventricular diastole immediately before LA contraction [164]. Phasic LA function was assessed with the calculation of LA empty-ing volumes and fractions: LA total emptyempty-ing volume and fraction (LAEV to-tal and LAEFtotal, corresponding to LA reservoir function), LA passive

CARDIOVASCULAR MAGNETIC RESONANCE FEATURE TRACKING FOR ASSESSING MYOCARDIAL MECHANICS: FROM EXPERIMENTAL STUDIES TO CLINICAL RESEARCH

Tomas Lapinskas

CARDIOVASCULAR MAGNETIC

RESONANCE FEATURE TRACKING

FOR ASSESSING MYOCARDIAL

MECHANICS:

FROM EXPERIMENTAL STUDIES

TO CLINICAL RESEARCH

Tomas Lapinskas

MIOKARDO DEFORMAVIMOSI RODIKLIŲ

ĮVERTINIMAS ŠIRDIES MAGNETINIO

REZONANSO TYRIMU

BANDOMUOSIUOSE IR

KLINIKINIUOSE MOKSLINIUOSE

TYRIMUOSE

CONTENTS

ABBREVIATIONS

INTRODUCTION

1. THE AIM AND OBJECTIVES OF THE STUDY

2. REVIEW OF LITERATURE

3. MATERIALS AND METHODS